Stochastic local search combined with LSB technique for image steganography

In this paper, we propose a steganography technique to hide a message in an image. The proposed technique combines the stochastic local search meta-heuristic (SLS) with the least significant bits method (LSB). First, we developed the basic technique of LSB to hide information in JPEG images. We found that LSB steganography method is not effective because the size of the image increases greatly after inserting information which facilitates the discovery of hidden information. In order to improve the performance, we added a meta-heuristic approach to LSB technique. We hybridized local search (LS) with LSB. Then we improved LSB by combining it with a stochastic local search (SLS). The three methods LSB, LSB + LS and LSB+ SLS are implemented and evaluated on some series of JPEG images. We decomposed the image and the secret message in 4-bits blocks. Then by applying the principle of each method, we are able to insert the secret message in the images. A key is generated and encrypted with the AES algorithm. The numerical results are interesting and demonstrate the benefits of our proposed method.